The two principal motivations of our neotectonic study of Myanmar have been to understand the past occurrences of and future potential for large earthquakes in the Myanmar region.
Throughout the previous pages, we have accomplished the former by constructing a new
neotectonic map that helps to make sense of many of the large earthquakes of the past century or so.
Looking to the future, public seismic safety will depend to a large extent on understanding the potential for other large earthquakes throughout this region. Our neotectonic map assists in this goal, as well. Many of the active faults within the region have not produced large earthquakes during the past century or more of human record-keeping, so what is their potential for the future?
Although what we have presented is by no means complete, our geomorphologic mapping augmented by seismic, geodetic and other relevant geological data, provides a fundamental basis for a simple evaluation of the predominant seismic sources for each of the three active tectonic domains of Myanmar and its neighboring countries. In this section we utilize our neotectonic understanding of these active faults to assess their potential for future rupture. The current scarcity of published structural information, high-quality seismological and geodetic data and
paleoseismological information limits this effort to a pretty basic level. Nonetheless, we provide below a synoptic, first-order estimate of plausible earthquake scenarios within each domain.
Wells and Coppersmith (1994) (W&C) provide equations that relate rupture length to earthquake magnitude. Blaser et al., (2010) improved upon these scaling relationships by using a enlarged historical earthquake database. They also incorporated thrust fault ruptures in subduction environments, thus enabling better estimates of earthquake magnitude for such faults.
Length-magnitude scaling relationships for subduction megathrusts have also been given by Strasser et al. (2010). These two independent scaling relationships help us to estimate the
uncertainties in estimation of maximum earthquake magnitude produced by the megathrust along the western Myanmar coast. Table 3 lists the parameters that we used to calculate potential earthquake magnitude.
Although we used structural discontinuities, jogs and kinks to define structural segments and assumed that these segment boundaries delimit plausible future fault ruptures, we are well aware that fault ruptures sometimes propagate through such structural complications and thus produce larger earthquakes (e.g., Wesnousky, 2006). Currently, however, paleoseismological and historical documentation of rupture lengths in the Myanmar region are too sparse to warrant a sophisticated consideration of multiple–segment ruptures. In this first effort, we simply estimate magnitudes associated with single-segment rupture for the mapped faults. In some cases, we also use estimates of fault slip rate and published geodetic analyses to offer plausible average
earthquake recurrence-times for these full-segment ruptures. These simplistic average recurrence intervals provide a useful starting point for future hazard analyses.
Table 4 summarizes the potential earthquake magnitudes we have calculated for all of the major structures. Below, we explain these results for the faults of each domain, starting in the west with the four domains of the Indoburman range and ending in the east with the Shan-Sino domain.
The Indoburman range
We will assess the seismic potential of the four domains of the Indoburman range from south to north, in the same order that we described them in the preceding section. In addition to the surface manifestation of these domains that appear in the maps of Figures 3a, 4a, 6 and 9a, we utilize four schematic cross-sections (Fig. 18 and 19), based upon available geological and seismicity data. Together the maps and cross-sections allow us to estimate the preliminary three-dimensional geometry of the megathrust and its relationship to large secondary structures. At this stage of our understanding of the geometries and kinematics of the region’s active faults, it seems unwarranted to conduct a statistical analysis of plausible rupture areas, widths and slip amounts. We attempt here merely a crude first cut at assessing earthquake potential of the region.
So for example, we do not attempt to include the range of uncertainty in the depth of down-dip
rupture limits for the megathrust; instead, we mainly use the length of fault mapped from the surface to assess the plausible maximum earthquake magnitude on the subduction zone interface.
Coco-Delta domain
We have described above a Coco-Delta domain dominated by a highly oblique plate interface that dips about 20° to 30° eastward (Dasgupta et al., 2003) (Fig. 18a). The orientation of this section of the megathrust (early parallel to the vector of relative plate motion), its steep dip and secondary features imply predominantly right-lateral slip across this oblique-reverse fault. A predominance of dextral slip within the domain, on the very northern part of the 2004 megathrust rupture (Chlieh et al., 2007), is consistent with this interpretation. The down-dip limit of its seismic rupture is likely shallower than the ~50-km down-dip limit of the adjacent megathrust farther south (Chlieh et al., 2007; Heurent et al., 2011), as its motion contains a large component of strike slip. However, the down-dip limit of the locked patch may still extend to about 20 or even 30 km, as the subducting oceanic lithosphere here is old and cold (> 80 Ma; Müller et al., 1997). We use the reverse fault and megathrust equations from both Blaser et al. (2010) and Strasser et al. (2010) of to estimate a Mw 8.6 to 8.9 range for the maximum earthquake that could be produced by this 480-km long segment (Table 4).
The fact that the southernmost part of this domain ruptured during the great 2004 earthquake (Meltzner et al., 2006), supports the suspicion that this section of the megathrust can accumulate tectonic strain and slip seismically. Moreover, large submarine landslides mapped by Nielsen et al., (2004) within this domain could well be evidence that the megathrust has produced high ground accelerations in the past. However, a complete rupture of the Coco domain megathrust segment would be very rare, because the ten or more meters of slip during such an event would take a millennium or longer to accumulate at the average slip rate of the fault, which could well be lower than 1 cm/yr.
In addition to the megathrust, we suggest that at least three other structures along the eastern flank of the Indoburman range may be capable of generating significant earthquakes (Table 4; Fig.
2 and Fig. 20). From their lengths, we estimate that Mw 7.6 to 7.7 earthquakes are plausible.
Lacking reliable estimations of their fault slip rate, however, it would be speculative to estimate average return periods of such earthquakes.
Ramree domain
The Arakan earthquake of 1762 may represent the maximum earthquake within the Ramree domain, because it appears to have resulted from failure of the megathrust in combination with large splay faults in the upper plate (Wang et al., 2013a). If the megathrust ruptured across the entire length of the domain, from near Fouls Island to Chittagong (Figure 4), as Cummins (2007) suggests, then the magnitude would likely have been within the range Mw 8.5 and 8.8, based on the average coseismic fault slip on the megathrust fault plane (Wang et al., 2013a). This range of magnitudes is consistent with our estimation of the maximum earthquake magnitude based upon fault length (Blaser et al., 2010 and Strasser et al., 2010; Table 4).
Terrace- and coral-uplift records yield recurrence intervals ranging between about 400 to 1000 years for earthquakes that involve uplift of Ramree and Cheduba Islands (Shishikura et al., 2009;
Wang et al., 2013a). This range is about twice as long as the 190- or 550-year recurrence interval calculated for Mw 8.6 to 8.8 earthquakes if the 23 mm/yr oblique plate convergence is fully taken up by slip on the 450-km long megathrust. This discrepancy implies either that the megathrust is not fully coupled or that the oblique Indian-Burman plate motion is partitioned between the megathrust and upper plate faults, such as the Thahtay Chaung fault within the Indoburman Range.
Upper-plate structures may fail separately from the megathrust and generate smaller, but nonetheless destructive earthquakes along the western Myanmar coast. The 1848 earthquake of
northern Ramree Island (Oldham, 1883) may be one of these events. It caused moderate damage to the city of Kyaukpyu, but the felt area was much more limited than that of the 1762 earthquake.
The great length of the right-lateral strike-slip Thahtay Chaung fault, within the Indoburman Range (Fig. 4a and 5) implies that this fault could generate the earthquake as large as Mw 7.6 (Wells and Coppersmith, 1994; Blaser et al., 2010). The lack of reliable written history in this mountainous region precludes knowing whether such an event has happened within the past 250 years. Moreover, a lack of constraints on the slip rate of the fault precludes us from saying anything meaningful about an average recurrence interval. Nevertheless, the existence of this large strike-slip fault within the Indoburman range gives good reason to hypothesize that large, destructive shallow earthquakes are plausible within the range.
The lengths of the west-dipping East Limb faults that crop out along the eastern flank of the Indoburman Range (Fig. 4a and 18b) imply that they are capable of generating Mw 7.8 and 7.3 earthquakes (Table 4). It is likely that these two faults may be connected in the subsurface by a blind thrust. If so, combined rupture could generate an even greater earthquake. The average recurrence interval of such an event along the eastern Indoburman Range would be greater than a thousand years, though, as GPS analysis shows the shortening rate across the eastern Indoburman Range and the central Burma basin is < 9 mm/yr (Socquet et al., 2006).
Several active reverse faults between Thayet-Myo and Yangon could generate large earthquakes along the floodplain of the Ayerawaddy River. Within the Ramree domain, the southernmost of these is the West Bogo-Yoma fault, on the eastern flank of the Ayerawaddy flood plain. The fault is likely a high-angle reverse fault that dips northeastward beneath the western flank of the Bago-Yoma Range. The length of the western Bogo-Yoma fault implies a maximum magnitude of Mw 7.2 to 7.3 for earthquakes near the Ayerwaddy flood plain north of Yangon. The Paungde fault, farther north along the Ayerwaddy flood plain, is longer, so we estimate that it is
capable of producing a Mw 7.3 to 7.4 earthquake in the vicinity of Prome. A related fault farther north likely produced the earthquake of 1858 and may have disrupted temporarily the flow of the Ayeyarwady River.
Before we leave the discussion of the Ramree domain, we should also note that rupture of faults within the downgoing slab could also produce damaging earthquakes in the region. Such hidden faults would not be manifest in our mapping of surface features, so we can say little more than that the existence of these should be contemplated in making a comprehensive seismic assessment of the region. The mb 6.5 Bagan earthquake of 1975 was an event of this type. Its hypocentral depth was about 120 km (Engdahl and Villasenor, 2002). The earthquake ruined several temples in the ancient capital of Burma that are believed to have been built in about the 12th century.
Dhaka domain.
The Dhaka domain is defined by the length and width of the broad belt of folds of the Chittagong-Tripura fold belt. As such it extends nearly 600 km along strike from south to north and more than 200 km from west to east. If the blind megathrust underlying this entire domain were to fail at once, the resulting earthquake would likely have a magnitude of about Mw 8.9 (Blaser et al., 2010; Table. 4).
Whether such a large event is plausible is currently a matter of some debate. Recent GPS studies above the megathrust show that the Indoburman Range is moving westward at least 5 mm/yr relative to the Indian plate (Steckler et al., 2012; Gahalaut et al., 2013). Whether this E-W shortening is reflects aseismic creep on or strain accumulation across the megathrust remains unclear. Gahalaut et al. (2013) argue the seismic risk from the underlying plate interface event is low because the E-W shortening on the N-S running megathrust is so low and they find no earthquakes in the history were sourced from the plate-interface. Steckler et al. (2008) argue from
comparison with other megathrusts with sediment-rich accretionary prisms that this section of the megathrust may well be capable of producing large earthquakes, even through the prism’s internal strength and basal friction are weak.
The lack of large historical earthquakes in the past 400 to 500 years for this portion of the megathrust does not mean the risk of such large megathrust event is low. In fact, if elastic E-W shortening is indeed only 5 mm/yr, it would take nearly 1000 years to accumulate enough slip potency for an Mw 8.9 earthquake on a fully coupled 520-km long and 350-km wide megathrust. If the megathrust is semi-coupled, as many megathrusts are (e.g. Chlieh et al., 2008; Hsu et al., 2012), the recurrence interval of such events would be even longer.
Regardless of whether the megathrust/décollement is capable of producing a giant earthquake, many upper-plate structures associated with actively growing young anticlines (Fig. 6) are undoubtedly capable of producing earthquakes, either in association with failure of the megathrust or individually. Using the lengths of young anticlines as indicators of the lengths of the underlying faults, we calculate plausible maximum earthquake magnitudes ranging from Mw 6.3 to 7.7 (Table 4). The 1918 Mw 7.5 earthquake near the Rashidpur anticline may be an example of such an event.
The 1999 mb 5.2 earthquake near the Maheshkhali anticline may be an example of partial failure of one of these faults.
Several other moderate but destructive earthquakes have struck within the fold belt during the pre-instrumental historical period. From the records of shaking alone, however, one cannot be certain that these were produced by failure of secondary structures above the megathrust. They could also have been caused by rupture of faults within the descending plate, beneath the décollement. Speculating about the recurrence intervals of these earthquake sources is not particularly useful because so little is known about the rate of slip on these structures or how their ruptures relate to ruptures of the subjacent megathrust/decollement.
Ruptures of faults within the down-going Indian Ocean lithosphere farther east are another plausible source of destructive earthquakes. One example is the Ms 7.4 earthquake of 1954, which struck east of the Indoburman Range. Its hypocentral depth is 180 km (Engdashi and Villasenor, 2002), clearly within the Wadati-Benioff zone of the downgoing slab. Fortunately, sources deeper than about 50 km within the Wadati-Benioff zone pose relatively low seismic hazard, because such ruptures are far from human infrastructure at the Earth’s surface. Shallower sources, however, within the subducting Indian Ocean lithosphere west of the crest of the Indoburman range, could cause destructive earthquakes within the populated regions of Bangladesh. Destructive earthquakes in Bangladesh in 1842 and 1885, for example, are reasonable intraslab candidates, as there is no geomorphic evidence of surface deformation near their proposed epicenters.
As in the Ramree domain to the south, the Dhaka domain has seismic faults within and east of the high Indoburman range. The Churachandpur-Mao fault is the most prominent of these. Judging by its 170-km length, wholesale failure of this right-lateral fault could produce an Mw 7.6 earthquake. The geodetic analysis of Gahalaut et al. (2013) suggests, however, the fault may be slipping aseismically. Aseismic slip along active strike-slip faults is usually associated with minor to moderate earthquakes (Lienkaemper et al., 1991). However, we did not find any historical events that could be related to the Churachandpur-Mao fault in the earthquake catalog of Szeliga et al.
(2010), nor does the instrument catalog show a high level of seismic activity along the fault.
The ~280-km length of the eastward dipping Kabaw fault implies that it could generate an Mw 8.4 earthquake if it were to fail all at once (Table 4). The average interval between such earthquakes would be a millennium or longer, since geodetic analysis suggests the fault slip rate must be lower than 9 mm/yr (Socquet et al., 2006).
Naga domain
The southeastward-dipping Naga thrust fault is the principal seismic source within the Naga domain (Fig 9). The 400-km length of the fault implies a maximum earthquake of Mw 8.5 to 8.7 (Table 4). The structural cross section from Kent (2002) suggests the dip of the Naga thrust fault is about 23°, higher than the dip angle of the megathrust of the Dhaka and Ramree domains. In addition to being distinguished by a steeper dip, the fault is also distinguished by the fact that it is the interface between two pieces of continental lithospheres (Fig. 19b; Fig. 1a), rather than the convergent boundary between oceanic lithosphere and the continental lithosphere.
Using the equations from Blaser et al. (2010) and Strasser et al. (2010), we estimate that the Naga thrust fault is capable of producing an Mw 8.5 to 8.7 earthquake, similar in size to the great Assam earthquake of 1950, which resulted from rupture of the Himalayan Frontal Thrust, just to the north. On the other hand, it is plausible that each of the three 100- to 150-km long arcuate lobes we have mapped commonly fail individually. In such a case, the magnitude of the largest Naga thrust earthquakes would be in the Mw 7.7 to Mw 8 range.
The slip rate of the Naga thrust fault is constrained neither by GPS vectors spanning the fault nor by vectors from plate-motion models. GPS vectors on either side of the western part of the Naga thrust are similar, so it appears that there is no shortening across this thrust fault system (Jade et al., 2007; Maurin et al., 2010). This ostensibly conflicts with recent field investigations that show the Naga thrust overrides Quaternary alluvium at the mountain front (Aier et al., 2011). Clearly, additional work will be needed to resolve this important question about the seismic potential of the fault.
Although we did not find any evidence of active faults along the southeastern flank of or within the northern Indoburman range (Fig. 9a), some intraslab earthquakes occur within the down-going Indian plate beneath the range. Although the intraslab events in the Naga domain were
not as frequent as those in the Dhaka domain, failure of faults within the subducting plate poses a potential hazard within the Naga domain. In addition, several earthquakes greater than magnitude 6 have occurred along the southeastern margin of the northern Indoburman range. These include a magnitude 7 earthquake in 1932 and an Mw 7.2 earthquake in 1988 (Engdashi and Villasenor, 2002). Both of these two events originated at depths greater than 90 km. Although their hypocenters are deep, the 1988 earthquake still caused some damages to the nearly regions.